Information & Management 47 (2010) 226–236

Contents lists available at ScienceDirect

Information & Management

journal homepage: www.elsevier.com/locate/im

To give or to receive? Factors influencing members’ knowledge sharing and

community promotion in professional virtual communities
Chih-Jou Chen a, Shiu-Wan Hung b,*
a
Department of Information Management, National Penghu University, Penghu, Taiwan
b
Department of Business Administration, National Central University, No. 300, Jung-Da Road, Jung-Li City, Tao-Yuan 320, Taiwan

A R T I C L E I N F O A B S T R A C T

Article history: Professional virtual communities (PVCs) bring together geographically dispersed, like-minded people to
Received 4 February 2008 form a network for knowledge exchange. To promote knowledge sharing, it is important to know why
Received in revised form 21 December 2009 individuals choose to give or to receive knowledge with other community members. We identified
Accepted 11 March 2010
factors that were considered influential in increasing community knowledge transfer and examined
Available online 27 March 2010
their impact in PVCs. Data collected from 323 members of two communities were used in our structural
equation modeling (SEM). The results suggested that norm of reciprocity, interpersonal trust, knowledge
Keywords:
sharing self-efficacy, and perceived relative advantage were significant in affecting knowledge sharing
Knowledge contributing
behaviors in PVCs. The knowledge contributing and collecting behaviors were positively related to
Knowledge collecting
Virtual communities knowledge utilization. Furthermore, while the collecting behavior had a significant effect on community
Individual factors promotion, the influence of contributing behavior on community promotion was limited.
Contextual factors ß 2010 Elsevier B.V. All rights reserved.

1. Introduction important to know why individuals choose to give or to receive

The emergence of professional virtual communities (PVCs) has
provided a solution, new insights, and mechanisms into knowledge
management and sharing in organizations. VCs are groups of
people who communicate regularly in an organized manner via a
communication medium, such as a bulletin board or a news group.
Rapid exchange of information and knowledge through VCs has
dramatically changed lifestyles, enhancing individual and organi-
zational learning. Community members share their ideas and
thoughts. Individuals take part in PVCs to secure knowledge,
resolve problems, improve individual capability, absorb specia-
lized knowledge, and create innovations. Many organizations have
recognized PVCs as valuable in their knowledge management and
have begun to support their development and growth [1].
The process of knowledge sharing involves members con-
tributing knowledge and seeking knowledge for reuse. Success of
PVCs requires knowledge contributors to be amenable to donating
their knowledge and be willing to reuse codified knowledge [2].
The same individual can, of course, be a giver or a receiver at
different times. The biggest challenge in facilitating a VC is the
continuous supply of knowledge from members. It is thus
* Corresponding author. Tel.: +886 3 4269903; fax: +886 3 2118558.
E-mail address: shiuwan@mgt.ncu.edu.tw (S.-W. Hung).
knowledge when they have a choice.
PVC members use virtual meeting places to discuss opportu-
nities for new products and product improvement. Therefore,
identifying the critical factors underlying knowledge sharing
behavior would help in understanding knowledge sharing. Despite
various attempts to examine the behavior of knowledge sharing
[3–9], none has adequately explained why individuals choose to
give or receive knowledge when they have a choice.
We developed an integrated model to examine the effects of
contextual factors, including norm of reciprocity and interpersonal
trust, and individual factors, including knowledge sharing self-
efficacy, perceived relative advantage, and perceived compatibility,
on knowledge utilization and community promotion through
knowledge sharing. Prior research indicated that knowledge sharing
was facilitated by a strong sense of reciprocity and fairness [10].
Interpersonal trust is a key need for effective collaboration [11], and
is the salient factor in determining the effectiveness of knowledge-
sharing activities. On the other hand, some researchers suggested
that individual motivators may affect knowledge contributors’
willingness to share knowledge [12]. The expectation of individual
benefits will encourage members to share knowledge. Therefore,
this research has attempted an integrated model that explores the
effectiveness of knowledge sharing from both contextual and
personal perspectives. Data collected from 323 members of two
PVCs were collected for our study. The analysis used commercially
available LISREL 8.7 software.

0378-7206/$ – see front matter ß 2010 Elsevier B.V. All rights reserved.
doi:10.1016/j.im.2010.03.001
 C.-J. Chen, S.-W. Hung / Information & Management 47 (2010) 226–236
2. Theoretical background and hypothesis
2.1. Virtual communities
The VC is concerned with social interaction among its members.
We focused on the PVC that consists of people with common
interests in specialized fields or subjects for knowledge sharing.
They provide a knowledge communication channel allowing
sharers to interact via the Internet: a form of computer-mediated
communication [13]. However, knowledge management studies
have discovered that its availability is no guarantee that knowl-
edge sharing will take place: social interactive issues have
significant effect.
2.2. Knowledge sharing and influencing factors
Recently, researchers have defined factors that affect an
individual’s willingness to share knowledge: its costs and benefits,
incentive systems, extrinsic and intrinsic motivation, social capital,
social and personal cognition, organization climate, and manage-
ment championship [14,15]. Most previous studies have focused
on either contextual factors or on personal factors in knowledge
sharing.
Social Cognitive Theory (SCT) argues that a person’s behavior is
partially shaped and controlled by the influences of contextual
factors and the person’s cognition; it states that an individual will
take an action that has personal cognition in a social environment.
According to SCT, the question ‘‘Why do individuals choose to
give to or receive knowledge from other community members?’’,
should be addressed from a perspective of both contextual factors
and personal cognition. To investigate the knowledge sharing
behavior in PVCs, we used SCT to develop a research model where
Fig. 1. An integrated framework for examining knowledge sharing.
227
contextual, personal, and behavioral factors act as interacting
determinants that influence one another bidirectionally. We
focused on the role of contextual and individual factors influencing
personal behavior. Social Exchange Theory explains human
behavior in social exchanges, which differ from economic
exchanges because the obligations are not clearly specified. In
these exchanges, people do others a favor with a general
expectation of a future return. Therefore, social exchange assumes
the existence of a relatively long-term relationship, as opposed to a
one-off exchange. The norm of reciprocity and interpersonal trust
are two of the most significant contextual factors that drive
knowledge sharing. Its self-efficacy, perceived relative advantage,
and compatibility are seen as predictors of personal factors, since
they are all considered to be the main influences shaping user
behavior [16,17].
In our study, we proposed an integrated framework to develop a
more comprehensive perspective of the relationships between
contextual factors, personal factors, knowledge sharing behavior,
and community promotion, and bring it up-to-date with empirical
data from two PVCs. The integrated analytic framework is shown in
Fig. 1.
2.3. Norm of reciprocity
The norm of reciprocity usually refers to a set of socially
accepted rules regarding a transaction in which a party extending a
resource to another obligates the latter to return the favor [18], and
has been highlighted as a benefit for individuals engaging in social
exchange. A basic norm of reciprocity is a sense of mutual
indebtedness, so that individuals reciprocate by repaying the
benefits they receive from others, ensuring ongoing supportive
exchange. Reciprocity can serve as a motivational mechanism to
228 C.-J. Chen, S.-W. Hung / Information & Management 47 (2010) 226–236
contribute to discretionary databases. In our study, norm of
reciprocity refers to people’s salient belief that current knowledge
sharing will lead to a future request for knowledge being met.
Generally speaking, participation in VCs is open and voluntary
and participants are not acquainted. Knowledge seekers have no
control over who responds to their question or its quality.
Knowledge contributors have no assurance that those they are
helping will ever return the favor. This sharply contrasts with
traditional communities of practice and face-to-face knowledge
exchanges where people typically know each other and interacts
over time, creating expectations of obligation and reciprocity
enforceable through social sanctions. The norm of reciprocity
represents a pattern of behavior where people respond to friendly
or hostile actions with similar actions. Wasko and Faraj suggested
that people who share knowledge in online communities believe in
reciprocity. In the collective climate, reciprocity is the norm that
facilitates the sharing of knowledge. This lead to:
Hypothesis 1a. The norm of reciprocity is positively related to the
knowledge contributing behavior of members in PVCs.
Hypothesis 1b. The norm of reciprocity is negatively related to the
knowledge collecting behavior of members in PVCs.
2.4. Interpersonal trust
Trust is, of course, a necessity in knowledge sharing on the
Internet [19–22]. In general, trust develops when a history of
favorable past interactions leads to positive expectations of future
interaction. Interpersonal trust in others’ abilities, benevolence,
and integrity increases the desire to give and receive information,
resulting in improved performance of distributed groups; it creates
and maintains exchange relationships. Trust is a good predictor of
e-commerce adoption [23,24], business cooperative relationships,
and performance [25].
In our study, interpersonal trust implies a degree of belief in
good intentions, benevolence, competence, and reliability of
members who share knowledge in PVCs. It resides in the whole
community. Facilitating cooperation and knowledge exchange
[26]. Thus, when there is strong interpersonal trust, individuals
will liberally exchange, seek, and collect knowledge in VCs. Lee and
Choi [27] examined lack of trust among employees as a major
barrier in knowledge-sharing activities. When there is high trust in
employee relationships, employees become more willing to
participate in knowledge-sharing [28]. These arguments lead to:
Hypothesis 2a. Interpersonal trust is positively related to the
knowledge contributing behavior of members in PVCs.
Hypothesis 2b. Interpersonal trust is positively related to the
knowledge collecting behavior of members in PVCs.
2.5. Knowledge sharing self-efficacy
Individuals attempting to improve others’ perception of their
competency are motivated by internal self-efficacy, which is a form
of self-evaluation that influences a person’s decision on behaviors
to undertake, the amount of effort and persistence to excel when
faced with obstacles, and mastery of the behavior. In general,
perceived self-efficacy plays an important role in influencing
individuals’ motivation and behavior. Therefore, people who have
high self-efficacy will be more likely to perform related behaviors
than those with low self-efficacy.
Knowledge sharing self-efficacy is one’s confidence in an ability
to provide knowledge that is valuable to others. In our study, this is
the member’s self-evaluation and confidence in his or her skills and
capabilities to respond to questions posted by other members, and
to provide knowledge that is valuable and useful to others.
Through sharing useful knowledge, people feel more confident in
what they can do. Contributing knowledge to others helps enhance
VC users’ learning and self-efficacy. Hence:
Hypothesis 3a. Knowledge sharing self-efficacy is positively
related to the knowledge contributing behavior of members in
PVCs.
Hypothesis 3b. Knowledge sharing self-efficacy is positively
related to the knowledge collecting behavior of members in PVCs.
2.6. Perceived relative advantage
Relative advantage is a measure of the degree to which an
action provides more benefit than its precursor. It is manifested as
increased efficiency and effectiveness, economic benefit, and
enhanced social status. In our study, perceived relative advantage
is the knowledge contributor’s cognition of likely advantages and
benefits that knowledge sharing behavior will produce and return
to him or her. Its value is positively related to the rate of adoption.
Correspondingly, facilitators of knowledge sharing have reported
obvious benefits, such as reduced communication costs and faster
problem-solving. In general, when decisions makers perceive clear
overall personal and organizational benefits of knowledge sharing,
they are more likely to encourage a knowledge-sharing culture in
the organization. Some studies suggested that individuals would
share knowledge within PVCs with a reward of enriching knowl-
edge, seeking support, making friends, etc. They are also seen as
skilled, knowledgeable or respected. Furthermore, people who
regularly helped others in PVCs seemed to receive help more
quickly when they asked for. These arguments lead to:
Hypothesis 4a. Perceived relative advantage is positively related
to the knowledge contributing behavior of members in PVCs.
Hypothesis 4b. Perceived relative advantage is positively related
to the knowledge collecting behavior of members in PVCs.
2.7. Perceived compatibility
Motivated individuals behave in congruity with their value
systems, irrespective of the driving force behind the motivation.
Compatibility is the degree to which an innovation fits into the
existing values, previous experience, and current needs of
potential adopters. In our study, perceived compatibility refers to
the knowledge contributor’s cognition of likely value, need, and
experience that his or her knowledge sharing behavior is similar to
the original value system. Moreover, greater fit is desirable,
because it can motivate individuals to develop new ideas. Thus,
members in PVCs perceiving knowledge sharing as compatible
with their individual values and needs are more likely to be
positively predisposed to adopting and promoting it; leading to:
Hypothesis 5a. Perceived compatibility is positively related to the
knowledge contributing behavior of members in PVCs.
Hypothesis 5b. Perceived compatibility is positively related to the
knowledge collecting behavior of members in PVCs.
2.8. Knowledge-sharing activity
PVCs can consist of multiple knowledge bases as well as the
mechanisms for collection, management, and utilization of the
knowledge. However, without a rich, diversified, and renewable
 C.-J. Chen, S.-W. Hung / Information & Management 47 (2010) 226–236
supply of knowledge, they are of limited value. Success requires
that knowledge contributors are willing to donate their knowledge
and that knowledge seekers are amenable to reusing the codified
knowledge. Clearly, effective PVC must receive a continuous
supply of knowledge. If a PVC embodies an ideal community
process of knowledge contributing and collecting, it will be likely
to lead its members to participate as well as to frequently visit the
bases. These ideas lead:
Hypothesis 6a. Members’ knowledge contributing behavior is
positively related to the knowledge utilization from PVCs.
Hypothesis 6b. Members’ knowledge collecting behavior is posi-
tively related to the knowledge utilization from PVCs.
PVC members can provide valuable knowledge for others or
reply to requests from help-seeking members. Knowledge sharing,
disseminating ideas quickly, and providing emotional support are
frequently observed in a PVC in the form of intensive postings and
viewings by its members. Bulter [29] indicated that intensive
knowledge postings/viewings or frequent online interactions have
the potential to support higher level help-giving behavior and
social support. Thus, when members aggressively participate in
more knowledge-sharing activities in PVCs, they are more likely to
promote PVCs or invite new potential knowledge contributors.
Thus:
Hypothesis 7a. Members’ knowledge contributing behavior is
positively related to the community promotion.
Hypothesis 7b. Members’ knowledge collecting behavior is posi-
tively related to the community promotion.
According to the Social Cognitive Theory, individuals are more
likely to engage in behavior that they expect to result in favorable
consequences. Rich content and more personalized services
increase the participation of members in VCs. Since various
promotional services (e.g., technical forum, expert club, etc.) and
events (e.g., seminars, conferences, etc.) by PVCs are frequently
available to community members, it is likely that members with a
higher level of community participation show a more positive
attitude towards their PVC. Eventually, community members are
more proactive and aggressive in promoting the PVCs to new
knowledge contributors. This leads to:
Hypothesis 8. Members’ knowledge utilization from PVCs is posi-
tively related to the community promotion.
3. Method and data
3.1. Respondents and procedure
An online (web-based) survey was conducted by sending it to
individual members of VCs. Several steps were taken to ensure
data validity and reliability by refining and rigorously pre-testing
the questionnaire. The questionnaire was pre-tested by four
experts in the information system area and five executives in
prominent PVCs. The comments collected from these people
provided a basis for the revisions of the construct measures and the
modifications of the wording and item sequence. Furthermore, two
pilot studies were completed. The first was conducted by 73
members in several knowledge discussion forums hosted by
PChome Online (www.pchome.com.tw), while the second was
completed by several PVCs members, including three professors,
15 PhD candidates, and 20 master degree candidates. These pilot
tests were intended to evaluate the relevance of the measurement
items to the research variables. Therefore, we used previously
229
validated measurement items wherever possible to help ensure
the validity of our survey; multiple-item measures were used for
most constructs to enhance content coverage. All of our multiple-
item constructs achieved Cronbach alphas of 0.75 or higher,
indicating strong internal consistency. In addition, we performed a
CFA on the measurement items using the samples of the two pilots.
All the factor loadings were significant and had high R2 values,
confirming convergent validity. Comments and suggestions on the
item contents and structure of the instrument were solicited.
Two of the largest IT-oriented PVCs in Taiwan were selected for
this analysis: the Programmer-Club community (www.program-
mer-club.com) with 170,000 members and the BlueShop commu-
nity (www.blueshop.com.tw) with 190,000 members. They both
dedicated to sharing knowledge in programming language,
databases, technical problems, network skills, and operating
systems, and are members of the Microsoft Community Alliance
Program. The research model was then tested with data received
from members of these PVCs. A banner with a hyperlink to our
online survey was posted on the homepage of the Programmer-
Club from April 20–May 31, 2007, and on the homepage of the
BlueShop from May 1–June 15, 2007. Members of these two PVCs
with knowledge sharing experience were invited to support this
survey. Twenty randomly selected respondents were offered an
incentive in the form of gifts. All respondents were informed that
the data collection was part of a research study on knowledge
sharing behavior. Additionally, to minimize potential confusion, a
definition and description of knowledge sharing was included at
the beginning of the questionnaire. A summary of the survey
results was also made available to respondents who requested it.
By the time this survey was concluded, 1282 visitors browsed the
survey, of which, 354 questionnaires were received. The exclusion of
31 invalid questionnaires resulted in a total of 323 complete surveys
for data analysis. Prior research has showed that participants from
self-selected samples provider clearer, more complete responses
than those who are not volunteers. Of course, the validity of any
research methodology relying on volunteers is contingent on their
ability and willingness to provide a meaningful response. The
counter in our survey website could tally the number of visits and
compute the response rate automatically. We deleted incomplete
returns even those with only a few items unanswered. The response
rate of 27.6% from 323 people was enough to generate statistically
reliable estimates on casual paths among constructs.
Non-response bias was evaluated using the usual method:
testing for significant differences between early and late respon-
dents, with late respondents being considered surrogate for non-
respondents. To examine the representative of the participating
members and non-response bias, we performed MANOVA to
compare the early 110 respondents with the last 110 respondents
on all of the variables. The results suggested no significant
differences (p < 0.05), indicating that response bias was not present.
We also performed MANOVA to compare 209 IT-oriented technical
respondents with the other 114 respondents on all variables. The
results suggested no significant difference between the respondents
of these communities (p < 0.05). Detailed descriptive statistics of
the respondents’ characteristics are shown in Table 1.
3.2. Measures
All the measurement items in the study were adapted from
prior research including norm of reciprocity, interpersonal trust,
knowledge sharing self-efficacy, perceived relative advantage,
perceived compatibility, knowledge-sharing activity (including
knowledge contributing and collecting) [30], knowledge utiliza-
tion, and community promotion [31]. Table 2 lists definitions of the
constructs and related literature. For the measures, a seven-point
Likert-type scale (ranging from 1 = strongly disagree to
230 C.-J. Chen, S.-W. Hung / Information & Management 47 (2010) 226–236

Table 1 Table 3
Sample characteristics (the number of subjects = 323). Model fit results for measurement and structural models.

Demographic characteristics Frequency Percentage Fit index Measurement model Structural model Recommended value
2
Gender Male 266 82% x /d.f. 282/241 = 1.17 334/252 = 1.32 3.00
Female 57 18% GFI 0.94 0.92 0.90
AGFI 0.93 0.90 0.90
Age <21 60 19% NFI 0.97 0.96 0.90
21–30 103 32% CFI 0.99 0.99 0.95
31–40 95 29% SRMR 0.03 0.04 0.08
41–50 42 13% RMSEA 0.02 0.03 0.05
>50 23 7%

Education High school or below 33 10%

College (2 years) 64 20% 4.1. Analysis of the measurement model
University 140 43%
Graduate school 72 23%
PhD 14 4%
The measurement model was first evaluated and then
respecified to generate the ‘best fit’. We used LISREL 8.7 to
Working experience <1 year 43 13%
evaluate the properties of the measures, addressing latent
1–3 22 7%
3–5 41 13%
constructs via confirmatory factor analysis (CFA). The construct
5–10 115 36% unidimensionality, convergent, and discriminant validity of the
10–15 62 19% candidate items and constructs used in a study were also assessed
15–20 27 8% by employing latent variable constructs. The CFA results indicated
>20 13 4%
that the measurement model fitted the data, x2/d.f. = 1.17,
Job title IS manager 21 7% GFI = 0.94, AGFI = 0.93, NFI = 0.97, CFI = 0.99, SRMR = 0.03 and
Project manager 14 4% RMSEA = 0.02 as shown in Table 3. All the model-fit indices exceed
Programmer 44 14%
Software engineer 38 12%
the normal common acceptance levels, demonstrating that the
Hardware engineer 36 11% measurement model exhibited a good fit with the data collected.
Web application engineer 27 8% The convergent validity of the scales was verified by using normal
System engineer 29 9% three criteria: that all indicator loadings should be significant and
Students 45 14%
exceed 0.7, that the composite reliability should exceed 0.7, and that
Others 69 21%
the average variance extracted (AVE) by each construct should
Member history <3 months 28 8% exceed the variance due to measurement error for that construct
3–6 months 38 12%
(i.e., should exceed 0.50). The CFA results were consistent with the
6 months–1 year 42 13%
1 year–2 years 88 27% relationships expected between the measured items and their
2 years–3 years 54 17% respective constructs. All were significantly related, via t-tests, to
Over 3 years 73 23% their respective constructs. The factor loadings for all items except
Online history <1 year 13 4% CP1 exceeded the recommended level of 0.7. From Table 4, the
1–3 years 22 7% composite reliability of the constructs ranging from 0.77 to 0.91 all
3–5 years 43 13% exceeded the benchmark of 0.7. The AVE ranged from 0.53 to 1. Also,
5–8 years 89 28%
the internal consistency reliability to test unidimensionality was
8–12 years 77 24%
Over 12 years 79 24%
assessed by Cronbach’s a. The resulting a values ranged from 0.75 to
0.92 (see Table 4), which were above the acceptable threshold of
0.70. Additionally, we used Partial Least Squares (PLS) Graph Version
7 = strongly agree) was used. Appendix A lists all survey items used 3.00 to calculate cross-loadings of latent constructs. There were only
to measure each construct. four item loadings greater than 0.3 onto other factors. The result
indicated a good degree of unidimensionality for each construct.
4. Data analysis and results These are shown in Appendix B. thus all the conditions for
convergent validity and unidimensionality were met.
We used a two-step procedure for assessing the reliability and The discriminant validity of the scales was assessed by
validity of the measures before their use in the full model. determining that the square root of the AVE from the construct

Table 2
Definitions of constructs.

Construct (abbreviation) Definition

Norm of reciprocity (NR) People’s salient beliefs that knowledge sharing would lead to future request for knowledge being met
Interpersonal trust (ITR) The degree of belief in good intentions, benevolence, competence, and reliability of members sharing
knowledge in PVCs
Knowledge sharing self-efficacy (KSSE) The degree of member’s self-evaluation and confidence in their skills and capabilities in responding
questions posted by others, and to provide knowledge that is valuable and useful
Perceived relative advantage (PRA) The degree to which encouraging knowledge sharing was perceived to benefit the members
Perceived compatibility (PC) The degree to which encouraging knowledge sharing was perceived to fit the value system and current
needs of members
Knowledge contributing behavior (KCB1) The quantity of knowledge contributing based on the average volume of a member’s active knowledge
donating and posting per month
Knowledge collecting behavior (KCB2) The quantity of knowledge collecting based on the average volume of a member’s knowledge collecting
and viewing per month
Knowledge utilization (KU) The degree of usage of knowledge obtained from PVCs
Community promotion (CP) The degree to which promoting a virtual community is aggressive in inviting new members to join and
talk about the benefits of the PVC
 C.-J. Chen, S.-W. Hung / Information & Management 47 (2010) 226–236 231

Table 4
Confirmatory factor analysis results of the measurement model.

Construct Internal reliability Convergent validity

Item Cronbach a Factor loading (t-value) CR

Norm of reciprocity NR1 0.90 0.82 (18.5***) 0.88

NR2 0.86 (19.9***)
NR3 0.85 (19.6***)

Interpersonal trust ITR1 0.92 0.82 (19.0***) 0.90

ITR2 0.91 (22.2***)
ITR3 0.87 (20.7***)

Knowledge sharing self-efficacy KSSE1 0.88 0.88 (21.4***) 0.91

KSSE2 0.89 (21.6***)
KSSE3 0.85 (20.1***)

Perceived relative advantage PRA1 0.86 0.77 (16.3***) 0.82

PRA2 0.79 (17.0***)
PRA3 0.76 (16.2***)

Perceived compatibility PC1 0.87 0.85 (20.0***) 0.89

PC2 0.85 (19.9***)
PC3 0.78 (17.6***)
PC4 0.81 (18.4***)

Knowledge contributing behavior KCB1 n/a 1.00 (27.7***) n/a

Knowledge collecting behavior KCB2 n/a 1.00 (27.7***) n/a

***
Knowledge utilization KU1 0.75 0.70 (13.9 ) 0.77
KU2 0.71 (14.1***)
KU3 0.76 (15.3***)

Community promotion CP1 0.85 0.62 (12.6***) 0.84

CP2 0.77 (16.7***)
CP3 0.80 (17.5***)
CP4 0.80 (17.6***)

Notes: (1) CR = composite reliability; AVE = average variance extracted. (2) See Appendix A for the description of each construct and item.
***
p < 0.001.

was greater than the correlation shared between the construct and
other constructs in the model. Table 5 shows the correlations
among the constructs, with the square root of the AVE on the
diagonal. All the diagonal values exceeded the correlations
between any pair of constructs, indicating the measure has
adequate discriminant validity.
Four correlations of independent variables were above 0.45 (see
Table 5), highlighting the need to test strictly for discriminant
validity. We therefore adopted another procedure to assess the
discriminant validity of the subsets of measures. Within each
subset, we examined pairs of constructs in a series of two-factor
confirmatory factor models, using LISREL. We ran each model
twice, once constraining the correlations between the two
constructs to unity and once freeing this parameter. Then a chi-
square different test was conducted. The results indicated that the
chi-square values were significantly lower for the unconstrained
models. In total, 20 models were run, involving 10 pairs of
comparison. The chi-square differences were all significant at
p < 0.001 (see Table 6), which suggested that the constructs
exhibited discriminant validity.
In addition we also checked for multicollinearity due to the
relatively high correlations among some constructs (e.g., a
correlation of 0.57 between PRA and KSSE). The resultant variance
inflation factor (VIF) values for all the constructs were acceptable
between 1.32 and 1.86. In summary, the measurement model
demonstrated adequate and sufficient reliability, unidimension-
ality, convergent validity, and discriminant validity.
Finally, we conducted a test to assess if common method bias
would be problem in interpreting the results. We used Harmon’s
single factor test, based on the assumption that if the risk of a

Table 5
Discriminant validity: correlations and AVE.

Construct Mean St. dev. AVE NR ITR KSSE PRA PC KCB1 KCB2 KU CP

NR 4.22 1.09 0.71 0.84

ITR 5.25 1.14 0.75 0.35** 0.87
KSSE 5.04 1.16 0.76 0.25 0.48** 0.87
PRA 4.70 1.08 0.60 0.12* 0.28** 0.57** 0.77
PC 4.13 0.99 0.68 0.34** 0.54** 0.49** 0.33** 0.82
KCB1 3.38 1.45 1.0 0.24** 0.45** 0.52** 0.55** 0.39** 1.0
KCB2 3.89 1.64 1.0 0.16** 0.55** 0.48** 0.58** 0.33** 0.41** 1.0
KU 4.88 0.93 0.53 0.07 0.17** 0.18** 0.20** 0.13* 0.27** 0.23** 0.73
CP 5.22 1.03 0.56 0.08 0.21** 0.20** 0.23** 0.14* 0.25** 0.31** 0.60** 0.75

Note: Diagonal elements are the square root of the average variance extracted (AVE). Off-diagonal elements are the correlations among constructs. For discriminant validity,
AVE should be larger than squared correlation between any pair of constructs; hence diagonal elements should be larger than off-diagonal elements.
*
p < 0.05.
**
p < 0.01 (two-tailed test).
232 C.-J. Chen, S.-W. Hung / Information & Management 47 (2010) 226–236

Table 6 tage, and perceived compatibility) and four latent endogenous

Discriminant validity analyses with pairs of constructs.
constructs (knowledge contributing behavior, knowledge collect-
Pairs of constructs Constrained Unconstrained Dx2 ing behavior, knowledge utilization and community promotion).
model x2 (d.f.) model x2 (d.f.) As shown in Table 3, all the model-fit indices exceeded their
NR with respective common acceptance levels: x2/d.f. = 1.32, GFI = 0.92,
ITR 162 (9) 32.4 (8) 130*** AGFI = 0.90, NFI = 0.96, CFI = 0.99, SRMR = 0.04 and RMSEA = 0.03,
KSSE 183 (9) 28.1 (8) 155*** suggesting that the model fitted the data well. The resulting
PRA 204 (9) 34.0 (8) 170***
structural model, together with the path coefficients and their
PC 247 (14) 27.8 (13) 220***
significance values is shown in Fig. 2. Eleven of the fifteen paths
ITR with exhibited at least a p-value less than 0.05, while the remaining four
KSSE 154 (9) 25.9 (8) 129***
PRA 175 (9) 28.6 (8) 146***
were not significant at the 0.05 level of significance.
PC 265 (14) 32.1 (13) 233*** As expected, norm of reciprocity had a negative direct effect on
knowledge collecting behavior (path coefficient = 0.11, p < 0.05)
KSSE with
PRA 192 (9) 35.2 (8) 157***
and no significant influence on knowledge contributing behavior.
PC 242 (14) 28.3 (13) 214*** Thus hypotheses H1b was supported while H1a was not.
Interpersonal trust was positively related to knowledge contribut-
PRA with
PC 277 (14) 38.7 (13) 239***
ing behavior (path coefficient = 0.13, p < 0.05), and had a relatively
***
strong positive direct effect on members’ knowledge collecting
p < 0.001.
behavior (path coefficient = 0.3, p < 0.001). Hence, hypotheses H2a
and H2b were supported. Knowledge sharing self-efficacy had a
common method is substantial, a single latent factor will explain quite strong positive direct effect on knowledge contributing
most of the covariance among the measures [32]. The single-factor behavior (path coefficient = 0.23, p < 0.001), and a positive direct
model resulted in x2(275) = 4655, compared with x2(241) = 282 effect on knowledge collecting behavior (path coefficient = 0.16,
for the measurement model, indicating that common method bias p < 0.05). Hence, hypotheses H3a and H3b were supported.
was not a serious threat in our study. Perceived relative advantage had quite a strong positive direct
effect on members’ knowledge contributing behavior (path
4.2. Test of the structural model coefficient = 0.32, p < 0.001), and on knowledge collecting beha-
vior (path coefficient = 0.36, p < 0.001). Hence, hypotheses H4a
The causal structure of the hypothesized research model and H4b were supported. Perceived compatibility had a weak
reflecting the assumed linear, causal relationships among the direct effect on knowledge contributing behavior (path coeffi-
constructs was tested for the model. The LISREL analysis used five cient = 0.12, p < 0.1) and no significant influence on members’
exogenous latent constructs (norm of reciprocity, interpersonal knowledge collecting behavior. Consequently, hypotheses H5a was
trust, knowledge sharing self-efficacy, perceived relative advan- supported while H5b was not.

Fig. 2. Results of structural equation model (SEM) analysis.

 C.-J. Chen, S.-W. Hung / Information & Management 47 (2010) 226–236
In terms of H6a and H7a, members’ knowledge contributing
behavior had a strong direct effect on knowledge utilization (path
coefficient = 0.22, p < 0.01) and no significant influence on
community promotion. Members’ knowledge collecting behavior
had a significant influence on knowledge utilization (path
coefficient = 0.14, p < 0.05) and on community promotion (path
coefficient = 0.18, p < 0.01). Hence, hypotheses H6b and H7b were
supported. Our results showed that knowledge utilization had a
strong significant influence on community promotion (path
coefficient = 0.55, p < 0.001), providing support for H8. The
explanatory power of the research model was depicted in Fig. 2.
5. Discussion
Our study provided an integrated model examining the
influence of both contextual and individual factors on knowledge
utilization and community promotion through knowledge sharing.
The R2 values show that contextual factors and individual factors
account for 40% of the variance in knowledge contributing
behavior and 44% of variance in knowledge collecting behavior.
Additionally, the knowledge sharing behaviors (contributing and
collecting) account for 29% of the variance in knowledge utilization
and 39% in community promotion.
5.1. Contextual factors
Surprisingly, the norm of reciprocity did not show a significant
influence on members’ knowledge contributing behavior. The
norm of reciprocity was found to be significant and has a negative
effect on members’ knowledge collecting behavior. The PVCs
sharply contrast with traditional communities of practice and face-
to-face knowledge exchanges.
Interpersonal trust significantly and positively influenced
members’ knowledge contributing and knowledge collecting
behaviors. They also show that trust has a significant effect on
the context of resource exchange and production innovation. It
was also found to have more influence on members’ knowledge
collecting than on contributing behavior. Nevertheless, the
forming, building and performing of interpersonal trust in PVCs
is slower and more fragile than that in organizations, due to a lack
of face-to-face settings and close and frequent interactions.
5.2. Individual factors
Knowledge sharing self-efficacy plays a vital role the knowl-
edge-sharing activities. It was found to influence members’
knowledge contributing and collecting behaviors; it also had a
more pronounced influence on members’ knowledge contributing
than collecting behavior.
Perceived relative advantage was found to be significant and
positively related to both the members’ knowledge contributing
and collecting behaviors. Furthermore, perceived relative advan-
tage had a rather similar and pronounced effect on members’
knowledge contributing and collecting behaviors.
Our model suggested that there were weak relationships
between perceived compatibility and members’ knowledge con-
tributing and collecting behaviors.
5.3. Knowledge sharing, utilization and community promotion
Our results suggested that the knowledge contributing and
collecting behaviors were positively related to knowledge utiliza-
tion. Furthermore, while the knowledge contributing behavior had
a significant effect on community promotion, the influence of
members’ knowledge contributing behavior on community
promotion was limited. An effective and beneficial knowledge
233
collecting activity may thus motivate members to invite their
friends to join a PVC. This implies that knowledge contributing has
an indirect but critical influence on community promotion through
knowledge utilization. Finally, knowledge utilization was found to
play a vital role underlying community promotion in reaching a
critical mass of community members within a short time, and this
brings the benefit of knowledge and network externalities to the
PVC.
6. Implications
Previous research has shown that VCs have significant
implications for strategic management of knowledge sharing by
its members.
The empirical result of our study suggested that both norm of
reciprocity and interpersonal trust are significant factors in
knowledge utilization and community promotion. Our finding
showed that the individual factors (knowledge sharing self-
efficacy, perceived relative advantage, and compatibility) should
be considered in order to make an appropriate decision for
selecting the knowledge-sharing activity as predicators of knowl-
edge utilization and community promotion in PVCs.
For practical applications, the analysis in this study offers a
framework of reference for the desire to facilitate knowledge
sharing (collecting and contributing), knowledge utilization and
community promotion within their organizations and virtual
knowledge communities or individuals initiating knowledge-
sharing practices. The results also suggest that interpersonal trust
plays an important role in knowledge utilization, community
promotion and knowledge-sharing activities within PVCs. Inter-
personal trust helps eliminate resistance barriers to knowledge
sharing, and without interpersonal trust, the cooperation required
for successful knowledge sharing might be difficult to attain or
nonexistent. For instance, managers can offer programs to enrich
knowledge and can organize social interaction to improve
members’ trust. In addition, managers can should attempt to
create better trusting relationships among employees by facilitat-
ing norm of reciprocity, sharing vision (experiences), dialoguing
and by confiding personal information in organizations or
communities, etc. Additionally, since knowledge sharing self-
efficacy, perceived relative advantage and compatibility are were
found to be important predictors of in motivating to smooth the
way knowledge is sharing, top management in organizations
should actively provide useful information and cases on the
benefits of knowledge sharing. The new information of knowledge
sharing programs gathered, together with the successful para-
digms for knowledge sharing from other organizations and
communities of practice can serve as positive examples of
knowledge sharing to employees (members).
7. Limitations
Our research had some limitations. First, the external validity of
the results may not extend beyond specially defined and selected
samples. The data was collected only at two websites of PVCs in
Taiwan. Second, the data presented were cross-sectional. The
development of contextual and individual factors leading to
knowledge-sharing activity, knowledge utilization and commu-
nity promotion is an ongoing phenomenon. These influencing
factors were measured at a static point rather than as they were
developing, thus losing time richness of explanation.
Third, since we simply measured the variable of knowledge
utilization and community promotion instead of ultimate out-
comes, the direct relationship between the level of knowledge
activity and its financial contribution to the PVC or the individual
was not explored.
234 C.-J. Chen, S.-W. Hung / Information & Management 47 (2010) 226–236

Appendix A. Measurement items used in the study

Norm of reciprocity (NR) (adapted from [12,33])

 I know that other members will help me, so it’s obligated and fair to help other members in this virtual community.
 When I share knowledge to other members, I believe that the members in this virtual community would help me if I need it.
 When I share knowledge to other members, I believe that my queries for knowledge well be answered in future in this virtual community.

Interpersonal trust (ITR) (adapted from [20,21,27,34,35])

 Members in this virtual community have reciprocal faith-based and trustworthy relationships.
 Members in this virtual community will not take advantage of others even when the profitable opportunity arises.
 Members in this virtual community will always keep the promise that make to one another.

Knowledge sharing self-efficacy (KSSE)

 I have confidence in my ability to provide knowledge that other members in this virtual community consider valuable.
 I have the expertise, experiences and insights needed to provide knowledge valuable for other members in this virtual community.
 I have confidence in responding or adding comments to messages or articles posted by other members in this virtual community.

Perceived relative advantage (PRA)

 Sharing knowledge with members in this virtual community will increase my solving- problem capability.
 Sharing knowledge with members in this virtual community will rapidly absorb and react to new information regarding the area.
 Sharing knowledge with members in this virtual community will be effective in my job and improve my performance.

Perceived compatibility (PC)

 Sharing knowledge with members in this virtual community is compatible with my values.
 Sharing knowledge with members in this virtual community fits my current needs.
 Sharing knowledge with members in this virtual community is compatible with my previous experiences.
 Sharing knowledge with members in this virtual community fits my work style.

Knowledge contributing behavior (KCB1)

Average volume of knowledge contributing per month (converted to seven-point scale)
& less than 2 times & about 2 to 5 times & about 6 to 10 times & about 11 to15 times
& about 16 to 20 times & about 21 to 30 times & more than 30 times

Knowledge collecting behavior (KCB2)

Average volume of knowledge collecting per month (converted to seven-point scale)
& less than 5 topics & about 5 to 10 topics & about 11 to 15 topics & about 16 to 20 topics
& about 21 to 30 topics & about 31 to 50 topics & more than 50 topics

Knowledge utilization (KU)

 I often use some kind of knowledge which I get from our virtual community to solve problems in my work.
 I frequently use some kind of knowledge which I get from our virtual community to improve professional knowledge level in my
expertise area.
 I regularly use some kind of knowledge which I get from our virtual community to handle challenges and changes of the work in the
future.

Community promotion (CP) (adapted from [6])

 I frequently talk to people about benefits of our virtual community.

 I usually spend some time to provide useful suggestions to our virtual community.
 I often introduce my peers or friends to our virtual community.
 I actively invite my close acquaintances to join our virtual community.
 C.-J. Chen, S.-W. Hung / Information & Management 47 (2010) 226–236 235

Appendix B. Item loading and cross-loading

Construct Items 1 2 3 4 5 6 7

Norm of reciprocity NR1 0.83 0.13 0.08 0.13 0.27 0.13 0.26
NR2 0.87 0.14 0.11 0.15 0.23 0.22 0.29
NR3 0.85 0.13 0.12 0.09 0.19 0.16 0.10

Interpersonal trust ITR1 0.23 0.82 0.20 0.08 0.23 0.11 0.08
ITR2 0.32 0.92 0.04 0.25 0.17 0.17 0.28
ITR3 0.23 0.87 0.26 0.22 0.19 0.11 0.09

Knowledge sharing self-efficacy KSSE1 0.21 0.20 0.87 0.20 0.21 0.13 0.22
KSSE2 0.11 0.29 0.90 0.20 0.11 0.28 0.11
KSSE3 0.15 0.10 0.85 0.31 0.26 0.25 0.16

Perceived relative advantage PRA1 0.20 0.24 0.30 0.77 0.34 0.14 0.16
PRA2 0.22 0.13 0.31 0.80 0.19 0.13 0.11
PRA3 0.11 0.08 0.21 0.76 0.23 0.07 0.25

Perceived compatibility PC1 0.16 0.12 0.25 0.19 0.85 0.08 0.23
PC2 0.26 0.11 0.20 0.22 0.86 0.15 0.19
PC3 0.19 0.00 0.12 0.18 0.79 0.23 0.25
PC4 0.11 0.13 0.22 0.17 0.81 0.27 0.24

Knowledge utilization KU1 0.07 0.12 0.17 0.12 0.17 0.71 0.05
KU2 0.12 0.13 0.10 0.23 0.17 0.72 0.13
KU3 0.14 0.06 0.20 0.20 0.05 0.75 0.20

Community promotion CP1 0.19 0.11 0.17 0.21 0.12 0.28 0.63
CP2 0.14 0.12 0.13 0.13 0.16 0.13 0.78
CP3 0.21 0.08 0.28 0.14 0.15 0.11 0.81
CP4 0.09 0.20 0.23 0.18 0.04 0.12 0.81

References
[1] P. Gongla, C.R. Rizzuto, Evolving communities of practice: IBM global services
experience, IBM Systems Journal 40 (4), 2001, pp. 842–862.
[2] S. Ba, J. Stallaert, A.B. Whinston, Introducing a third dimension in information
systems design: the case for incentive alignment, Information System Research 12
(3), 2001, pp. 225–239.
[3] C.M. Chiu, M.H. Hsu, E.T.G. Wang, Understanding knowledge sharing in virtual
communities: an integration of social capital and social cognitive theories,
Decision Support Systems 42 (3), 2006, pp. 1872–1888.
[4] H. Hall, D. Graham, Creation and recreation: motivating collaboration to generate
knowledge capital in online communities, International Journal of Information
Management 24 (3), 2004, pp. 235–246.
[5] M.H. Hsu, T.L. Ju, C.H. Yen, C.M. Chang, Knowledge sharing behavior in virtual
communities: the relationship between trust, self-efficacy, and outcome expecta-
tions, International Journal of Human-Computer Studies 65, 2007, pp. 153–169.
[6] J. Koh, Y.G. Kim, Knowledge sharing in virtual communities: an e-business
perspective, Expert Systems with Applications 26 (2), 2004, pp. 155–166.
[7] M. Ma, R. Agarwal, Through a glass darkly: information technology design,
identity verification, and knowledge contribution in Online communities, Infor-
mation Systems Research 18 (1), 2007, pp. 42–67.
[8] G. Widen-Wulff, M. Ginman, Explaining knowledge sharing in organizations
through the dimensions of social capital, Journal of Information Science 30 (5),
2004, pp. 448–458.
[9] C.P. Yu, T.H. Chu, Exploring knowledge contribution from an OCB perspective,
Information & Management 44 (3), 2007, pp. 321–331.
[10] M.M. Wasko, S. Faraj, It is what one does: why people participate and help others
in electronic communities of practice, Journal of Strategic Information Systems 9
(2–3), 2000, pp. 155–173.
[11] P.A. Pavlou, D. Gefen, Building effective online marketplaces with institution-
based trust, Information Systems Research 15 (1), 2004, pp. 35–62.
[12] A. Kankanhalli, B.C.Y. Tan, K.K. Wei, Contributing knowledge to electronic knowl-
edge repositories: an empirical investigation, MIS Quarterly 29 (1), 2005, pp. 113–
143.
[13] R.M. Wachter, N.D.J. Gupta, M.A. Quaddus, IT takes a village: virtual communities
in support of education, International Journal of Information Management 20 (6),
2000, pp. 473–489.
[14] G.W. Bock, Y.G. Kim, Breaking the myths of rewards: an exploratory study of
attitudes about knowledge sharing, Information Resource Management Journal
15 (2), 2002, pp. 14–21.
[15] G.W. Bock, R.W. Zmud, Y.G. Kim, J.N. Lee, Behavioral intention information in
knowledge sharing: examining the roles of extrinsic motivators, social-psycho-
logical forces, and organizational climate, MIS Quarterly 29 (1), 2005, pp. 87–111.
[16] C.L. Sia, H.H. Teo, B.C.Y. Tan, K.K. Wei, Effects of environmental uncertainty on
organizational intention to adopt distributed work arrangements, IEEE Transac-
tion on Engineering Management 51 (3), 2004, pp. 253–267.
[17] P.C. Verhoef, F. Langerak, Possible determinants of consumers’ adoption of
electronic grocery shopping in the Netherlands, Journal of Retailing and Con-
sumer Services 8 (5), 2001, pp. 275–285.
[18] J.B. Wu, P.W. Hom, L.E. Tetrick, L.M. Shore, L. Jia, C. Li, L.J. Song, The norm of
reciprocity: scale development and validation in the Chinese context, Manage-
ment and Organization Review 2 (3), 2006, pp. 377–402.
[19] W.S. Chow, L.S. Chan, Social network, social trust and shared goals in organiza-
tional knowledge sharing, Information & Management 45 (7), 2008, pp. 458–465.
[20] C.L. Hsu, J.C.C. Lin, Acceptance of blog usage: the roles of technology acceptance,
social influence and knowledge sharing motivation, Information & Management
45 (1), 2008, pp. 65–74.
[21] H.F. Lin, Determinants of successful virtual communities: contributions from
system characteristics and social factors, Information & Management 45 (8),
2008, pp. 522–527.
[22] J.B. Thatcher, M.L. Loughry, J. Lin, D.H. McKnight, Internet anxiety: an empirical
study of the effects of personality, belief, and social support, Information &
Management 44 (4), 2007, pp. 353–363.
[23] D. Gefen, E. Karahanna, D.W. Straub, Trust and TAM in online shopping: an
integrated model, MIS Quarterly 27 (1), 2003, pp. 51–90.
[24] P.A. Pavlou, M. Fygenson, Understanding and predicting electronic commerce
adoption: an extension of the theory of planned behavior, MIS Quarterly 30 (1),
2006, pp. 115–143.
[25] D.L. Paul, R.R. McDaniel Jr., A field study of the effect of interpersonal trust on
virtual collaborative relationship performance, MIS Quarterly 28 (2), 2004, pp.
183–227.
[26] P.S. Adler, Market, hierarchy and trust; the knowledge economy and the future of
capitalism, Organization Science 12 (2), 2001, pp. 215–235.
[27] H. Lee, B. Choi, Knowledge management enablers, processes, and organizational
performance: an integrative view and empirical examination, Journal of Manage-
ment Information System 20 (1), 2003, pp. 179–228.
[28] L.C. Abrams, R. Cross, E. Lesser, D.Z. Levin, Nurturing interpersonal trust in
knowledge-sharing networks, Academy of Management Executive 17 (4),
2003, pp. 64–77.
[29] B.S. Bulter, Membership size, community activity, and sustainability: a resource-
based model of online social structures, Information Systems Research 12 (4),
2001, pp. 346–362.
[30] B. Van den Hooff, F.D.L. Van Weenen, Knowledge sharing in context: the influence
of organizational commitment, communication climate and CMC use on knowl-
edge sharing, Journal of Knowledge Management 8 (6), 2004, pp. 117–130.
[31] S.S. Srinvasan, R. Anderson, P. Kishore, Customer loyalty in e-commerce: an
exploration of its antecedents and consequences, Journal of Retailing 78 (1),
2002, pp. 41–50.
[32] P.M. Podsakoff, S.B. MacKenzi, J.Y. Lee, N.P. Podsakoff, Common method biases in
behavioral research: a critical review of the literature and recommended reme-
dies, Journal of Applied Psychology 88 (5), 2003, pp. 879–903.
[33] M.M. Wasko, S. Faraj, Why should I share? Examining social capital and knowl-
edge contribution in electronic networks of practice MIS Quarterly 29 (1), 2005,
pp. 35–58.
[34] P. Palvia, The role of trust in e-commerce relational exchange: a unified model,
Information & Management 46 (4), 2009, pp. 213–220.
[35] C.M. Ridings, D. Gefen, B. Arinze, Some antecedents and effects of trust in virtual
communities, Journal of Strategic Information Systems 11, 2002, pp. 271–295.
236 C.-J. Chen, S.-W. Hung / Information & Management 47 (2010) 226–236
Chih-Jou Chen is currently the Assistant Professor at
the Department of Information Management of
National Penghu University, Taiwan. He received his
PhD from the Department of Business Administration of
National Central University, Taiwan. Before this, he
earned his Master Degree from National Central
University and his Bachelor degree from National Chiao
Tung University, Taiwan. His current research interests
focus on the development of virtual communities,
knowledge management, new product innovation and
performance management. His researches have been
published in Computers in Human Behavior, Interna-
tional Journal of Organizational Analysis, and other leading refereed journals and
conference proceedings.
Shiu-Wan Hung is currently the Associate Professor
at the Department of Business Administration of
National Central University, Taiwan. She received her
PhD from the Institute of Business and Management
of National Chiao Tung University, Taiwan. Before
this, Dr. Hung earned her MS from the University of
Wisconsin-Madison in USA and completed her BS
from National Chengchi University of Taiwan. Her
researches mainly focus on areas of knowledge
management, technology management and perfor-
mance management. Dr. Hung has had articles
published in the Scientometrics, Technovation, Com-
puters in Human Behavior, European Journal of Operational Research and other
journals.